It is not clear where the accent originates. One fallacy is that it has its origins in the Midwestern accent. Rather the accent of the upper Midwest is distinct and quite a departure from General American English.
The Telsur Project of William Labov and others examines a number of phonetic properties by which regional accents of the U.S. may be identified. The area with Midwestern regional properties is indicated on the map: eastern Nebraska (including Omaha and Lincoln), southern and central Iowa (including Des Moines), and western Illinois (including Peoria and the Quad Cities but not the Chicago area).
Since the 1960s northeastern Ohio and much of the rest of the Inland North have been affected by the Northern Cities Vowel Shift.
"The fact that the NCS is well established in Michigan is particularly interesting in light of the dominant beliefs about local speech. As research by Dennis Preston has shown, Michiganders believe they are “blessed” with a high degree of linguistic security; when surveyed, they rate their own speech as more correct and more pleasant than that of even their fellow Mid-westerners. By contrast Indianans tend to rate the speech of their state on par with that Illinois, Ohio, and Michigan. Indeed, it is not uncommon to find Michiganders who will claim that the speech of national broadcasters is modeled on their dialect. Even a cursory comparison of the speech of the network news anchors with that of the local news anchors in Detroit will reveal the fallacy of such claims.
The phoneme /ʍ/ is present only in varieties that have not undergone the wine-whine merger. /ʍ/ is often analyzed as a consonant cluster of /hw/. Also, many Americans realize the phoneme /ɹ/ (often transcribed as /r/) as a retroflex approximant [ɻ].
Depending on one's analysis, people who merge the vowels of cot and caught to /ɑ/ either have no phoneme /ɔ/ at all or have the [ɔ] only before /r/. Words like north and horse are usually transcribed /nɔɹθ/ and /hɔɹs/, but since all accents with cot and caught merged to /kɑt/ have also undergone the horse-hoarse merger, it may be preferable to transcribe north and horse /noɹθ/ and /hoɹs/. Thus, in these cases, the [ɔ] before /ɹ/ can be analyzed as an allophone of /o/. Some speakers who have maintained the contrast between /ɑ/ and /ɔ/ realize /ɔ/ phonetically lower, closer to [ɒ].
[ɝ] and [ɚ] are often analyzed as sequences of , respectively. [ə] is actually an indeterminate vowel that occurs only in unstressed syllables. Since the occurrence of [ə] is mostly predictable, it need not be considered a phoneme distinct from /ʌ/.
The diphthongs of General American are shown in the next table:
|Diphthongs||Offglide is a front vowel||Offglide is a back vowel|
|Opener component is unrounded||aɪ eɪ||aʊ|
|Opener component is rounded||ɔɪ||oʊ|
One phenomenon apparently unique to General American is the behavior of words that in RP have [ɒɹV] where [V] stands for any vowel. These words are treated differently in different North American accents: in New York-New Jersey English they are all pronounced with [-ɑɹ-] and in Canadian English they are all pronounced with [-ɔɹ-] (thus "sorry" is pronounced by Canadians as "sore-ee"). But in General American there is a split: the majority of these words have [-ɔɹ-], like Canadian English, but the last four words of the list below have [-ɑɹ-], like New York-New Jersey English, for many speakers. Words of this class include, among others: